The role of the native language in the use of the English nongeneric definite article by L2 learners: A cross-linguistic comparison

Abstract

The study uses an elicited imitation (EI) task to examine the effect of the native language on the use of the English nongeneric definite article by highly proficient first-language (L1) Spanish and Russian speakers and to test the hierarchy of article difficulty first proposed by Liu and Gleason (2002). Our findings suggest that there is a clear influence of L1 on participants’ reproduction of the second-language (L2) definite article in nongeneric contexts, but that various contexts present different levels of difficulty for the two L1 groups. The participants whose L1 is Spanish – a language with an article system – perform at a native-like level of accuracy in the grammatical condition of the test, whereas the participants whose L1 is Russian – a language without articles – demonstrate a tendency to omit definite articles in the same contexts. In the ungrammatical condition, Spanish speakers differ from the native speaker control group in their suppliance of the definite article in conventional and cultural contexts, while Russian participants supply the definite article significantly less than both the Spanish participants and the control group along all article categories. The study offers novel insights into what constitutes article difficulty for L2 learners from different L1s.

Keywords

definite article elicited imitation English hierarchy of article difficulty implicit knowledge L1 transfer nongeneric article Spanish Russian

I Introduction

The role of transfer of linguistic knowledge from the native language (L1) to a second language (L2) has been one of the central issues in the course of second language acquisition (SLA) research (for a detailed discussion of different aspects of L1 transfer, see Ellis, 1994; Gass and Selinker, 1992; Kellerman, 1995; Kellerman and Sharwood-Smith, 1986; Odlin, 1989; Papadopoulou and Clahsen, 2003; Ringbom, 1987, 1990; Schwartz and Sprouse, 1996, among many others). The explanation of the cognitive mechanisms underlying such transfer dates back to the behaviorists’ ideas of habit formation and cumulative learning. It is believed that learning in general, and language learning specifically, relies on the ability to use information from previously encountered tasks to enhance the learning of new tasks, provided that the tasks are similar in nature. As applied to SLA, the relevance of prior linguistic knowledge depends on the relationship that can be established between the learner’s L1 and L2. Because languages can differ in a variety of independent ways, it is more appropriate to establish such cross-linguistic relationships at different levels of linguistic analysis, e.g. phonological, morphological, etc. Consequently, the presence of similar (in form and/or function) linguistic elements in the L1 should facilitate acquisition of these elements in an L2 (positive transfer) because people naturally look for similarities and confirmation of their hypotheses before they look for differences (Schachter, 1983). On the other hand, if L1 and L2 elements are substantially different along some relevant parameters of a given L2 feature, this may create an interfering effect (negative transfer), which manifests itself in persistent and systematic errors. In addition to similarity and contrast relations between the L1 and the L2, there may exist a zero relation, i.e. when a given L2 feature does not even have a distantly similar counterpart in the learner’s L1. In such a case, learners cannot relate certain L2 elements to their previous linguistic knowledge and may lack the concepts necessary to perceive the fundamental distinctions in the L2 (Ringbom, 2007). Consequently, such learners may have more difficulty in learning these elements, or may altogether fail to make use of these elements in the L2, thus falling short of native-like mastery of the language.

While the effects of contrast and zero relation between the learners’ L1 and L2 can be revealed in the form of a negative transfer or avoidance strategies, and can be approached from the perspective of error analysis (Corder, 1967, 1981), the effect of a positive transfer is not immediately visible. Suppose that an L2 learner has acquired a certain L2 structure successfully, should we attribute this success to the similarity of the form and/or the function of that structure in the L1? How can we make sure that this effect is not due to a myriad of other possible factors? One approach to control for the extraneous factors as the cause of the observed L2 behavior is to compare performance of otherwise matched, comparable L2 learners from those L1s that stand in a different relation to the learners’ common L2 (Jarvis, 2000).

The present study adopts such comparative approach and examines the acquisition of an L2 feature by two groups of language learners: those learners whose L1 does not have a similar feature and those whose L1 does. Specifically, the goal of this study is to test and compare the use of the English definite article by highly proficient adult native speakers of Russian (a language that does not have articles) and Spanish (a language with a complex article system). Although such a combination of languages has been used in the past to examine acquisition of L2 articles by speakers of ‘article’ versus ‘article-less’ languages, the present study is novel and original in two principled ways.

We know from anecdotal and experimental data that L2 speakers do not produce articles equally poorly or equally well in all contexts because some uses of articles present a greater difficulty than others. By capitalizing on previous findings about article difficulty (García Mayo, 2008; Liu and Gleason, 2002), this study intends to examine the role of the native language in the use of the English definite article in its nongeneric interpretation by L2 learners from different L1 backgrounds. Because Spanish and English are similar with regard to the nongeneric interpretation of the definite article, L1 Spanish speakers are hypothesized to transfer the semantic features of articles from their L1 to their L2 in all nongeneric contexts and to outperform L1 Russian speakers of English because Russian lacks articles. If, however, we observe that for both groups of learners, Spanish and Russian, some nongeneric uses of the consistently elicit more erroneous responses than others, this will indicate that some article uses are inherently more difficult, and that the semantic transfer explanation alone cannot fully account for all the cases of L2 article difficulty.

Another important contribution of the present study is methodological. The study uses a modified oral elicited imitation procedure, specifically designed with the goal of tapping learners’ implicit knowledge of L2 article use. Although research on L1 transfer is plentiful, studies investigating L1 transfer effects with implicit measures are not many and are far from conclusive. However, in order to understand L2 learners’ difficulty with articles better, we need to examine not only what learners know explicitly and how they utilize their explicit knowledge, but also their behavior at the implicit level. The present study aims to fill the gap in the literature by examining how L2 learners from different L1 backgrounds utilize their implicit knowledge of different nongeneric uses of the L2 English article, and whether the observed difficulty with articles for speakers of an article-less L1 (compared to those of an L1 with articles) persists in an implicit task.

II Acquisition of L2 English articles

1 Article difficulty and complexity

Language instructors and educators unanimously agree that articles constitute the most difficult and complex structural elements for learning, and might even belong to the ‘unteachable’ grammar (Dulay et al., 1982). Articles are not perceptually salient in the input because they are unstressed (Master, 2002) and can be used variably even by native speakers of English depending on the pragmatic intent. But the main complexity of articles is attributed to the fact that they do not have a one-to-one form and meaning mapping (in English as well as in other article languages). Instead, multiple functions – e.g. noun countability, noun type (proper versus common), number, etc. – are stacked on a single morpheme (Butler, 2002; Master, 1997, 2002; White, 2009).

Studies that have compared article choice in English (indefinite a, definite the, and Ø) by L2 learners with different L1s show that, overall, the definite article emerges early and the indefinite article later in L2 acquisition, but that learners tend to omit articles in all article environments in the early stages of L2 acquisition, especially if they do not have articles in their L1 (Master 1997; Parrish, 1987; Thomas, 1989). Besides article omissions in obligatory contexts, learners sometimes substitute one article for the other (e.g. a for the, and vice versa) or oversupply articles in contexts where no article is required. Evidence of these types of errors is found in studies using different research approaches and methodologies examining article performance in L2 speakers from a variety of L1 backgrounds: Chinese (Leung, 2002; Robertson, 2000; Tryzna, 2009), Czech (Young, 1995), Finnish (Jarvis, 2002; Ringbom, 1987), Greek (Hawkins et al., 2006), Japanese (Butler, 2002; Hawkins et al., 2006; Mizuno, 1999; Snape et al., 2009), Korean (Ionin, 2003; Ionin et al., 2004), Polish (Ekiert, 2004; Tryzna, 2009), Russian (Ionin, 2003; Ionin et al., 2004, 2008), Serbian (Avery and Radišić, 2007; Trenkic, 2002, 2004, 2007), Spanish (García Mayo, 2009; Ionin and Montrul, 2010; Ionin et al., 2008; Montrul and Ionin, 2010; Snape et al., 2009), and Turkish (Goad and White, 2004, 2009; Snape et al., 2009), among many others.

Numerous studies have been carried out with the goal of identifying where such errors originate, and why such difficulty with articles persists despite plenty of language exposure and positive evidence, even in highly advanced and end-state L2 speakers. Such line of research has yielded plentiful data and rich evidence suggesting that multiple factors are at play.

Some researchers argue that the cause of the errors (at least some types of error) is related to a semantic parameter in Universal Grammar (UG), which is called the Article Choice Parameter (Ionin, 2003). According to this parameter, articles may encode either definiteness (e.g. in English and German) or specificity (e.g. in Samoan), but not both. The parameter of definiteness reflects the state of knowledge of both the speaker and the hearer, whereas the feature of specificity reflects the mind state of the speaker only, i.e. when the speaker singles out a unique subject or object and considers it to possess some noteworthy property (Ionin et al., 2004). Within the UG approach, it is argued that speakers of article-less L1s have access to the ACP parameter, but fluctuate between definiteness and specificity. This leads to errors in the use of L2 English articles because English only encodes definiteness (Hawkins et al., 2006; Ionin, 2003; Ionin et al., 2004). Some other approaches have sought explanations of article difficulty in the domain of phonology. For example, the Prosodic Transfer Hypothesis proposed by Goad et al. (2003) and Goad and White (2004) claims that L1 prosodic representations constrain production of L2 functional morphology, including articles. In other words, if the learner’s L1 does not permit certain kinds of prosodic representations required to produce an L2 structure, the learner will have difficulties in representing such morphology in his or her grammar. Morphosyntactic and syntactic explanations have also been put forth to explain learners’ difficulty with L2 articles. For example, the Syntactic Misanalysis Account (Trenkic, 2007) suggests that L1 speakers of languages that do not have a syntactic category of determiner (e.g. Serbian, Japanese) tend to miscategorize L2 determiners, including articles, as nominal modifiers. Thus, article omission errors are significantly more frequent in nouns adjectivally modified (Art + Adj + N) than non-adjectivally modified (Art + N) (Trenkic, 2002, 2004, 2007). However, despite several existing accounts of article acquisition difficulty, none offers an exhaustive and comprehensive explanation. Therefore, the question of L2 article acquisition remains as important and relevant as ever. In the present article, we continue examination of what constitutes article difficulty with a focus on just one aspect of the English article: its nongeneric interpretation.

2 Nongeneric use of the English article

Besides definiteness and indefiniteness, English articles can also encode genericity and nongenericity (Celce-Murcia and Larsen-Freeman, 1999; Hawkins, 1978; Master, 1997). When the definite article is used generically, it indicates a class of entities (e.g. a species, a race, or people of a nation). It is usually used with singular count nouns, although it may also be used with plural nouns, e.g. The German is very athletic, or The Germans are very athletic. All other uses of the are considered nongeneric. It is argued that the nongeneric use of the is much wider and more frequent than the generic one, and also more complex and hence more problematic for L2 learners (Liu and Gleason, 2002).

Different classifications of the nongeneric the have been proposed. Drawing on the early work of Christophersen (1939) and Jespersen (1949), Hawkins (1978) developed a comprehensive classification (the Location Theory) of the various nongeneric uses of the English definite article by grouping them into eight categories. More recently, Liu and Gleason (2002) revisited Hawkins’ taxonomy of the English article with the goal of investigating whether different nongeneric uses present the same learning difficulty for students of English as a second language (ESL) with different levels of proficiency. They combined the original eight nongeneric categories into four broad types: cultural, situational, structural, and textual (for the definitions and examples, see Table 1). Their participants – ESL learners from 18 different language backgrounds – were presented with an untimed, 91-sentence cloze test that contained 60 deleted obligatory uses of the (15 per each nongeneric article type) and were asked to supply articles where they deemed necessary.

Table 1.

Nongeneric uses of the (based on the classification of Liu and Gleason, 2002).

Article type		Definition	Example
1	Cultural	the is used with a noun that is a unique and well-known referent in a speech community	The Pacific Ocean is the largest ocean in the world.
2	Situational	the is used when the referent of a first-mention noun can be sensed directly or indirectly as unique by the interlocutors, or the referent is known by the members of the local community	While driving in their car to work, the husband asks his wife, ‘Could you open the window please?’
3	Textual	the is used anaphorically, with a noun that has been previously referred to or mentioned	Fendi bought a car on Monday. On Wednesday, he crashed the car.
4	Structural	the is used with a first-mention noun that has a modifier	The blue car across the road is very suspicious.

The results of Liu and Gleason’s study clearly indicated a hierarchy of difficulty of the four nongeneric uses. The cultural use turned out to be the most difficult one for all proficiency groups although low proficiency group performed significantly worse than intermediate and advanced groups. The situational the appeared to present the least difficulty out of the four nongeneric article types, but low proficiency group again scored significantly lower. The structural and the textual uses occupied an intermediate position between the situational and the cultural ones. The analysis of the learners’ underuse and overuse of the article revealed a clear developmental pattern: a correct suppliance of the article in obligatory contexts improved with proficiency while overuse of the article increased at a beginner level through intermediate level of proficiency but then decreased as proficiency improved from an intermediate to advanced level.

A series of subsequent, independently conducted studies employed the same methodology to investigate whether the proposed hierarchy of article difficulty holds true for specific L1 groups of English learners. Ansarin (2004) and Wong and Quek (2007) examined acquisition of the four nongeneric uses of the English definite article by native speakers of Persian, Mandarin and Malay, languages that do not have a functional equivalent of the English article system. García Mayo (2008) examined acquisition of the four nongeneric interpretations of the by L1 Spanish learners, speakers of a language with an article system. Interestingly, the three studies independently confirmed the hierarchy of difficulty hypothesis proposed by Liu and Gleason, both for those learners who have articles in their L1 and those who do not. While the cultural use of the nongeneric the presented the most challenge, the level of difficulty decreased for structural and textual uses, with the situational use being the easiest for all learners regardless of their native language.

Although the four studies examining L2 learners’ knowledge of the nongeneric the appear to have reached similar conclusions about the hierarchy of article difficulty, they all employed the same methodology and used the same instrument and materials. This creates a problem for generalizability of the findings because the results may reflect task or stimuli constraints instead of revealing a true hierarchy of article difficulty. The present study takes this issue into account and sets out to examine whether such hierarchy of difficulty can be validated using a different methodology, an elicited imitation task.

III The present study

1 Target L1s

This article adopts a cross-linguistic approach to examine the effect of learners’ native language (Spanish or Russian) on the use of different types of the English nongeneric definite article. Even though L2 acquisition of articles has been extensively researched over the past years and the given combination of L1s has already been examined, there has been no previous work directly comparing transfer effects in different L1 groups with regard to the use of the different nongeneric interpretations of the definite article.

In the study by Liu and Gleason (2002), the role of the native language was addressed in a post-hoc manner. They divided their ESL learners from 18 different language backgrounds into two broad groups: Indo-European and non-Indo-European. Given that English is an Indo-European language, they wanted to see if speakers of other Indo-European languages would make fewer errors than students from other language groups. They found significant differences only in the learners’ underuse of the in the cultural and situational categories and concluded that the native language was not a significant factor overall, at least not in all the usage types. Such comparison is obviously not well justified, and oversimplifies matters. There is a huge variation in terms of article use in Indo-European languages: while some Indo-European languages have articles (e.g. French, Spanish), others do not (e.g. Russian). Besides, simply because the learner’s L1 has articles, it does not mean that there will be an overlap in the use of all the four nongeneric functions of the definite article in English and the learner’s L1.

Ansarin (2004), García Mayo (2008) and Wong and Quek (2007) adopted a more stringent criterion for the sampling of their L2 participants, but each study focused on a single L1 group of English learners. García Mayo examined performance of L2 learners whose native language has articles, while Ansarin, as well as Wong and Quek, looked at performance of L2 speakers who come from article-less L1s. Differences in testing setting and proficiency measures between the three studies varied greatly, which makes it difficult to justify inter-group comparisons of the learners’ performance, and, thus, difficult to isolate the L1 effect. In the present study, we attempt to single out the L1 transfer effect by comparing two groups of L2 learners (L1 Russian and L1 Spanish) who are closely matched on L2 proficiency and language learning experience (such as years of formal instruction, age of arrival, age of onset, etc.).

a Russian

Russian does not have articles and does not grammatically encode definiteness or specificity (Ionin, 2003; Ionin et al., 2004). To illustrate, if we change the article in the same English sentence, we can change the sentence’s meaning, e.g. it happened on Monday (i.e. last Monday), it happened on a Monday (i.e. Monday and not any other day of the week), or it happened on the Monday (i.e. a particular Monday). A Russian speaker will not make a distinction between the three sentences and leave this to the context or the listener to clarify which Monday is intended. This is not to say that Russian does not have any means at all to indicate ±definite or ±specific referents. Although Russian does not grammaticalize (in)definiteness, this concept can be expressed at the lexical-semantic level with the help of pronouns, adjectives, adverbs and other parts of speech (Sidorova, 2006). For example, the numeral один (‘one’), or phrases какой-то (‘some’) or какой-нибудь (‘any’), may mark an indefinite reading. Definiteness, on the other hand, is most commonly indicated with the help of demonstrative pronouns, such as этот (‘this’) and тот (‘that’), as in (1a). These Russian demonstratives may also be optionally used in contexts where a textual nongeneric the (with nouns that have been previously mentioned), as in (1b), or a structural nongeneric the (with first-mention nouns that have modifiers), as in (1c), are used in English. However, similar to English, demonstrative pronouns in Russian cannot occur with cultural or conventional uses.

(1) a. Мне понравился (этот) фильм.

I liked-_PST (this)-_DEM movie-_MASC-ACC

‘I liked the/this movie.’

b. Я вчера посмотрел новый фильм. В (этом) фильме играют известные актёры.

I yesterday saw-_PST a new movie-_MASC-ACC. In (this)_-DEM movie-_MASC-PREP play famous actors

‘Yesterday, I saw a new movie. The/this movie features famous actors.’

с. Мне понравился (тот) фильм, который ты посоветовал.

I liked-_PST (that)-_DEM movie-_MASC-ACC which you recommended-_PST

‘I liked the/this movie which you had recommended.’

Definiteness and specificity may also be expressed at the syntactic level; for example, interact with word order (Christian, 1961; Ionin, 2003; Sidorova, 2006). For example, if a bare DP subject occurs pre-verbally, it can be interpreted as definite; if it occurs post-verbally, it can be interpreted as indefinite (Ionin, 2003). Moreover, the interpretation of the subject’s definiteness/indefiniteness is complicated by the fact that subjects in Russian do not have to precede the verb. With the majority of verbs it is normal for a subject to come first; but with certain types of verbs the reverse is true.

As can be seen from this brief overview, Russian uses some lexical and syntactic means (or a combination of both) to convey the meaning of (in)definiteness in certain contexts, but there is no direct one-way correspondence between the uses of the English article and the definite or indefinite subject interpretations in Russian.

b Spanish

English and Spanish have a substantial overlap in article semantics (i.e. what articles mean in the utterance context) and distributional properties (i.e. where articles can or must be used, and where none is required), but the two languages may vary in the use of articles in certain contexts, e.g. interpretation of articles in generic contexts (for a detailed discussion, see Ionin and Montrul, 2010; Montrul and Ionin, 2010; Snape et al., 2009). With regard to the nongeneric meaning of the definite article, the four nongeneric uses of the definite article in English seem to have their equivalent uses in Spanish (García Mayo, 2008; Leonetti, 1999; Vargas-Barón, 1952). Thus, cultural use includes a number of complex rules, as it does in English. For example, while names of countries do not normally take a definite article, there can be exceptions (compare Los Países Bajos in Spanish and the Netherlands in English, where a direct correspondence is observed). Similarly to English, names of rivers, seas and oceans, mountain ranges and archipelagos as well as unique and well-known referents in the speech community carry definite articles in Spanish. Definite articles in Spanish can also have situational, structural, and textual interpretations, as in (2a–c) (examples are taken from García Mayo, 2008):

(2) a. ¿Me pasas el mando a distancia de la tele?

to me pass-_SING-2nd the-_SING-MASC remote from the TV

‘Can you pass me the TV remote?’ (situational)

b. No nos gusta el novio que tiene ahora Elisa.

not-_NEG to us is pleasing-_SING-3rd the-_SING-MASC boyfriend whom has now Elisa

‘We don’t like the boyfriend Elisa has now.’ (structural)

c. Bryce llegó ayer a Santander. El conocido escritor participará en un curso de la Universidad en los próximos días.

Bryce arrived-_PST yesterday in Santander. The-_SING-MASC famous writer will participate in a course of the university in the following days

‘Bryce arrived in Santander yesterday. The well-known writer will participate in a university course in the following days.’(textual)

As can be seen, both English and Spanish require the use of the definite determiner in nongeneric contexts. Due to this overlap in article semantics, Spanish learners of English should be able to transfer their knowledge of article use from L1 to L2 across all nongeneric article types. However, there exists conflicting evidence that Spanish learners experience unequal difficulties with the different uses of the in English. According to García Mayo (2008), the cultural use of the definite article in English is the most difficult while the situational use is the easiest, mainly because beginner learners may not be familiar with actual cultural referents in the L2. It will be interesting to see, given a different set of materials and a different task, whether the same pattern of article difficulty will be observed, or whether Spanish speakers will perform similarly in all nongeneric contexts.

2 A methodological approach

We know that L2 learners’ performance and response accuracy vary under different task conditions (Ellis, 2000; García Mayo, 2007; Tarone and Parrish, 1989); that is why their knowledge of L2 structures needs to be probed with different procedures and methodologies. The present study is designed to examine L1 transfer effects with regard to the nongeneric interpretation of the English definite article in a more implicit, meaning-focus rather than form-focused task, an oral elicited imitation (EI) task.

The EI task is frequently used in research on child language acquisition (e.g. Ambridge and Pine, 2006; Eadie et al., 2002; Hirata-Edds, 2006) and L2 acquisition (Bley-Vroman and Chaudron, 1994; Ellis et al., 2006; Erlam, 2006; Graham et al., 2008) to determine the ability to use spontaneous, fluent, and contextualized language. The EI task has been claimed to tap test-takers’ implicit (as opposed to explicit) knowledge (Ellis, 2009; Erlam, 2006, 2009) because it (1) is time-pressured, (2) calls for a primary focus on meaning (not grammar), and (3) allows for automatic and parallel processing (rather than controlled processing) of language input without the participant’s awareness of how information is being processed. In addition, it also permits precise control over a specific target structure, a feature not available in other oral production methods, such as spontaneous production.

The mechanism behind the task is that it requires a test-taker to process a model sentence in several steps. The participant decodes the sentence through syntactic and semantic parsing, processes, retains the meaning, and reconstructs the sentence for subsequent production (Erlam, 2006). Because correct reproduction of a structure requires a person to have grammatical knowledge of that structure, it is hypothesized that people omit or substitute those parts of the input that are not represented in their knowledge system, i.e. structures that are beyond their current level of competence. The output utterance, therefore, is not a mere echoing of the original words, but a good reflection of whether a given structure has been integrated into a learner’s grammar or not (Erlam, 2006, 2009; Jensen and Vinther, 2003; Vinther, 2002).

Evidence in favor of the reconstructive nature of the EI task comes from a number of studies. In a study by Hamayan et al. (1977), Arabic learners of English were asked to repeat sentences with seven different types of grammatical structures. Half of the stimulus sentences were grammatically correct while the other half contained grammatical errors. Munnich et al. (1994) also presented learners with grammatical and ungrammatical sentences and asked them to simply repeat the sentences that they had heard. Both studies found that a significant number of learners ‘normalized’ sentences, i.e. corrected grammatical errors when they repeated the sentences. In yet another study, Markman et al. (1975) examined EI performance of L2 French learners and that of French native speakers and found that while learners did correct ungrammatical sentences, the native speakers made corrections significantly more often. A recent study by Granena (2009) compared participants’ rate of sentence correction in the EI task when test instructions explicitly asked them to correct ungrammatical sentences and when they did not. The results revealed that participants spontaneously corrected ungrammatical sentences even when they were not asked to do so.

The above evidence suggests that the EI task does not simply rely on rote repetition, but is reconstructive in nature. Therefore, participants’ correction or noncorrection of grammatical errors can be considered as an indication of their knowledge of L2 grammar. Using the EI task in the present study will allow us to evaluate L1 transfer effects with regard to learners’ more automatic, implicit system of knowledge. If L2 participants repair ungrammatical sentences containing article omission or misuse, this will signal that the knowledge of the L2 article is integrated in their knowledge system and that they are able to access and retrieve it automatically, routinely and spontaneously. If, however, they fail to repair ungrammatical sentences, this will indicate that the violations of article use do not interfere with meaning processing, and, consequently, the knowledge of the L2 article is not internalized.

IV Method

1 Participants

Forty-eight people participated in this study. Participants were divided into three groups (16 people per group) depending on their L1: English, Spanish and Russian (Table 2). The first group served as a control group and included monolingual speakers of American English who were born and raised in the USA. Most of them were undergraduate students, but several people were pursuing graduate degrees. Most Russian-speaking participants were born in Russia with the exception of two people: one who was born in Belarus and another who was born in the Ukraine, though both of them considered Russian their native language and spoke it from birth. The Russian participants had been living in the USA for an average of 3.63 years at the time of testing. Spanish participants came from different Spanish-speaking countries, including Argentina, Chile, Colombia, Costa Rica, Ecuador, El Salvador, Mexico and Venezuela. Spanish was the native language for all participants in the Spanish group; English was the second most proficient language after Spanish. Several participants also reported having beginner to intermediate proficiency in French and Portuguese.

Table 2.

Participants’ demographic profile.

Language group	Number of participants	Mean age	(Range)	Gender (M/F)
English	16	23.93	(18–57)	4/12
Russian	16	26.25	(20–34)	5/11
Spanish	16	31.75	(26–43)	11/5
Total	48	27.31	(18–57)	20/28

On average, the participants in the Spanish group had lived in the USA for 7.4 years. A Welch’s t-test was performed to compare the Russian and the Spanish participants’ mean length of residency in the USA. Using an alpha level of 0.05, the test showed that the difference in the length of residency between the two groups was not statistically significant, t(17.36) = −1.98, p = 0.064). Because English articles present a substantial learning difficulty for non-English speakers and are not mastered until a very late stage of second language development, only advanced L2 speakers of English were recruited to participate in this study. The Spanish and Russian participants in this study had on average 8.43 and 10.34 years of formal instruction in English before coming to the USA, respectively. A Welch’s t-test demonstrated that the group differences in length of formal instruction were not statistically significant, t(29.4) = 1.32, p = 0.198. Both Spanish and Russian participants reported a high percent of daily use of English, 67% and 78.13%, and rated their English proficiency as 7.97 and 8.16 on a 10-point scale, respectively. Table 3 summarizes the L2 speakers’ language learning background information.

Table 3.

L2 participants’ language background information (mean values).

L1	Age of onset in years	Duration of formal instruction in years	Age of arrival in years	Length of residence in years	Percentage of daily use of English	Overall self-rated proficiency*	Cloze test score
Spanish	9.93 (5.65)	8.43 (4.38)	24.19 (5.02)	7.4 (7.36)	67.00 (27.45)	7.97 (1.32)	41.88 (7.68)
Russian	9.00 (3.60)	10.34 (3.80)	21.25 (2.84)	3.63 (2.10)	78.13 (17.60)	8.16 (0.92)	43.81 (4.15)
Total	9.47	9.39	22.72	5.51	72.56	8.07	42.84

Note. * Lickert scale, where 1 is the lowest and 10 is the highest.

Besides the subjective self-ratings of overall language proficiency, Russian and Spanish speakers completed a 50-item English proficiency cloze test developed by Brown (1980). This test was used as part of the English Language Institute placement test at the University of Hawai’i, and is considered to be a valid and reliable measure of L2 learners’ vocabulary, morphosyntactic knowledge, and discourse competence. The maximum number of correct responses on the test is 50. Participants’ responses on the test were considered accurate as long as they were exact or acceptable. Each accurate – exact or acceptable – response received a score of 1, otherwise the response received a score of 0. These scores were added up to yield a final score indicating an individual L2 learner’s proficiency level. As can be seen in Table 3, Spanish participants scored 41.88 out of 50, and Russian participants scored 43.81 out of 50 possible points. A Welch’s t-test showed no between-group difference in mean cloze test score, t(23.08) = 0.89, p = 0.38, which suggests that both groups of L2 English speakers were comparable in terms of their L2 proficiency.

2 Procedure

The elicited imitation (EI) of oral production task was used in the present study. The procedure of this task is graphically presented in Figure 1. Attempts were made to make this task meaning-focused and reconstructive in order to tap participants’ automatic, implicit knowledge rather than gauge their ability to imitate given input verbatim. To this end, both verbal and nonverbal contextual support was provided, participants’ responses were timed, and sentence comprehension was measured.

Figure 1.

The procedure of the EI task: an example trial.

On each trial, participants saw a picture on the computer screen and a context sentence written under the picture. Next, participants heard the target sentence presented auditorily. The target sentence either correctly described the picture and matched the context sentence, or contained some sort of mismatch. Participants were required to make a decision about the target sentence by either agreeing with it or disagreeing with it based on the provided context, and then to repeat it after a short signal (30 milliseconds). Participants were instructed to start each response with the words ‘I agree that…’, or ‘I disagree that…’ depending on the preceding context. Half of the stimuli required an agreement response while the other half required a disagreement response. Participants were not informed that some of the target sentences were ungrammatical. Neither were they instructed to make corrections during the test. However, they were told at the beginning of the test that they may change the sentence if they do not remember the exact word order as long as meaning was preserved.

Immediately after the EI task, participants were debriefed and the immediate recall protocol was administered. The goal of the immediate recall protocol was to determine whether the participants were consciously aware of the errors in the EI task, what kind of errors they noticed, and whether they did any conscious corrections even though they were not instructed to do so. The recall protocol procedure included two rubrics, Noticing and Awareness of corrections, each consisting of three questions: one yes/no question, one Likert-scale question, and one open-ended question (Appendix 1).

3 Materials

The stimulus materials were adapted from the instrument used in the studies by Ansarin (2004), García Mayo (2008), Liu and Gleason (2002), and Wong and Quek (2007). In addition to the four categories of the nongeneric definite article identified in the above studies – cultural, textual, structural, and situational – one additional category was added. We refer to it as conventional use of the article the. It was added in order to make a distinction between the cases when the nongeneric definite article is used with proper nouns or well-known referents (e.g. the sun, the moon, the Capitol, the Pope, etc.), and the cases when its occurrence with certain words and expressions is quite systematic although not completely idiomatic (e.g. to play the piano, to take the train, to serve in the army, etc.). Both classes of these uses had been included in the cultural category in the previous studies. In all five critical nongeneric article conditions – cultural, textual, structural, situational and conventional – articles in the target positions occurred with singular non-adjectivally modified nouns (Art + N). Table 4 provides examples of test items used in the EI task. For each of the five categories of the article use, 10 sentence pairs were constructed resulting in a total of 50 critical test items. Each pair consisted of a context sentence (the sentence providing context for the article use) and a target sentence (the sentence where the use of the article was manipulated). Each target sentence came in two versions. One version contained the necessary article (grammatical condition), while the other omitted the necessary article (ungrammatical condition):

(3) Context sentence: The U.S. Congress consists of the Senate and the House of Representatives.

Target sentence:

a. grammatical: The Congress meets in the Capitol in Washington, D.C.

b. ungrammatical: The Congress meets in *Capitol in Washington, D.C.

Table 4.

Examples of critical test items in the elicited imitation task.

Article use	Context sentence	Target sentence
Cultural	The U.S. Congress consists of the Senate and the House of Representatives.	The Congress meets in the Capitol in Washington, D.C.
Conventional	There is a bad traffic jam, and I need to go to work.	I‘d prefer to take the train instead of driving.
Situational	This woman seems to be very happy with her new haircut.	She should give the hairdresser a generous tip.
Textual	Robinson Crusoe was cast away on a remote tropical island.	He had to live on the island for many years.
Structural	There is a beautiful ring with an emerald in the picture.	The ring is next to the watch with Mona Lisa.

Grammatical and ungrammatical items were counterbalanced between two presentation lists such that a grammatical version of a sentence occurred in list A while an ungrammatical version of the same sentence occurred in list B, and vice versa. This way, the same person was never exposed to the same sentence twice. This resulted in 25 grammatical and 25 ungrammatical critical test items in each presentation list. By comparing participants’ performance on both grammatically correct and incorrect sentences instead of only ungrammatical sentences, we should be able to see not only whether L2 speakers supply the necessary article when it is not present in the input, but also whether they successfully reproduce it when it is already present in the input.

In addition to the 50 sentences in the critical condition that involved manipulation of the obligatory use of the article, 20 (10 grammatical and 10 ungrammatical) control sentences were added. The control condition manipulated the use of the definite article with nouns that do not require any article at all such that while the ungrammatical sentences in the critical condition involved article omission in obligatory contexts, the ungrammatical sentences in the control condition involved overuse of the definite article:

(4) Context sentence: Mars is the fourth planet from the Sun.

Target sentence:

a. grammatical: There is no life on Mars.

b. ungrammatical: There is no life on *the Mars.

The control condition was included in order to prevent participants from developing a response bias in favor of consistently supplying articles with any noun they encounter. If participants decided to choose such a strategy, this would have yielded an inflated accuracy score in the critical test condition. Thus, having some sentences that did not require an article among those that did served as a control for bias.

Moreover, in order to prevent the participants from becoming too aware of the article violations in the test, 10 filler items were added to the stimulus materials to serve as a baseline condition. Grammatical violation in the filler condition concerned plural marking in nouns; for example:

(5) Context sentence: Smoking is very harmful for health.

Target sentence: A couple of * cigarette a day may eventually cause lung cancer.

It has been demonstrated before that, if the speakers’ L1 grammaticalizes plurality (e.g. Russian), they are more likely to notice violations in the use of plurals in the L2 than those speakers whose native language does not grammaticalize plurality (e.g. Japanese or Chinese) (Jiang et al., 2011). If our hypotheses about L1 transfer hold true and the instrument is sensitive to reveal test-takers grammatical competence, both Russian and Spanish participants are expected to autocorrect ungrammatical sentences in the filler condition to an extent comparable to native speakers because all these languages encode plurality grammatically. In contrast, because articles occur in Spanish and in English but not in Russian, Russian speakers of English are expected to accept ungrammatical sentences in the article condition as correct more often than Spanish speakers and English native speakers.

Finally, 10 practice sentences were constructed and included at the beginning of the test for familiarization with the task procedure. All practice sentences were grammatically correct because we did not want to heighten participants’ awareness of possible errors in the test. The structure of the stimulus materials is provided in Table 5.

Table 5.

The structure of the stimulus materials.

Condition	Grammatical	n	Ungrammatical	n
Critical	article present (+) article needed (+)	25	article not present (–) article needed (+)	25
Control	article not present (–) article not needed (–)	10	article present (+) article not needed (–)	10
Filler			violations of -s marking in plurals	10
Practice		10
Total		45		45

At the stage of materials development, the stimulus sentences were calibrated by a native speaker of English, a trained language instructor. She was given a list with the stimulus sentences from which all articles (both indefinite and definite) had been intentionally deleted, and was asked to supply ‘the’, ‘a(n)’, or both articles (if both were an acceptable option) where necessary. Based on the native speaker’s responses, only those items which unambiguously required the use of the in the test-critical position were included in the final test.

Care was taken to design the materials such that each target sentence was neither too long to be retained in memory nor too short to prevent rote repetition. On average, the sentences were about 13 syllables long. We also tried to control for the sentence structure and to rule out the potential occurrence of other kinds of determiners (e.g. possessive pronouns) in place of the definite article. The target sentences (both grammatical and ungrammatical) were recorded by a young female native speaker of American English three times and the clearest-sounding token was included in the test. The stimulus sentences were randomized to reduce the possibility of the inter-stimulus effects.

4 Results

The participants’ responses were recorded, transcribed and analysed. First, in order to ensure that the participants were attending to sentence meaning, their comprehension of the sentences was checked by analysing their accuracy of agreement/disagreement responses. Recall that in the EI task, participants had to agree or disagree whether the target sentence that they heard accurately matched the picture and the context sentence that they saw on the screen. As can be seen in Table 6, all participants in the three groups had high response accuracy scores, and the score difference between the three groups was not statistically significant, which is indicative of the fact that they were attending to the meaning of the sentences and comprehended them well.

Table 6.

Participants’ accuracy of agreement/disagreement responses.

Language	Sum of correct matches (out of 80)	SD	Percentage of correct matches	SD
English	73.50	1.67	91.88	2.09
Spanish	72.00	1.46	90.03	1.85
Russian	72.63	1.93	90.79	2.41

Next, the participants’ grammar correction behavior in the filler and the control conditions was analysed. These conditions were included in the instrument in order to prevent participants from developing response bias strategies and in order to provide a baseline comparison for the critical condition. In the filler condition, if the participants repaired the omission of plural -s, they received a score of 1, otherwise a score of 0. A cumulative elicited imitation (EI) score was calculated for each participant in each language group. A one-way analysis of variance (ANOVA) of the participants’ EI scores in the filler condition demonstrated that there was a significant between-participants main effect of the native language (F(2, 45) = 5.42, p < 0.01). A post-hoc Tukey HSD test however showed that there was no significant difference in the mean EI score of the Spanish (M = 8.6, SD = 1.6) and Russian (M = 7.8, SD = 1.9) speakers (p = 0.346). This suggests that both Russian and Spanish speakers were sensitive to the violations of the plural -s marking and corrected the errors most of the time although less frequently than the native speakers of English (Figure 2).

Figure 2.

Participants’ EI score in the filler condition.

For the control condition, a correctly omitted article in the target position received a score of 1; an incorrectly preserved article in the context when it is not needed received a score of 0. A two-way ANOVA of the participants’ EI score in the control condition yielded a significant between-participants main effect of the native language (F(2, 92) = 4.2, p < 0.05) and grammaticality (F(1, 92) = 66.56, p < 0.01). Although both native speakers and L2 speakers demonstrated a lower EI score in the ungrammatical condition compared to the grammatical condition, post-hoc Tukey HSD tests showed that Spanish and Russian groups repaired ungrammatical sentences involving article overuse even less often than the native speaker group, but that the mean EI scores were not significantly different in the thee language groups. These results are demonstrated in Figure 3.

Figure 3.

Participants’ EI score in the control condition.

Based on the analysis of the filler and the control conditions, it is evident that all participants showed a tendency to correct ungrammatical sentences even though they were not instructed to do so. Importantly, there were no significant differences between Russian and Spanish groups in terms of the frequency with which they made corrections. Therefore, if any performance differences are observed between the Russian and the Spanish groups in the critical condition, these differences should not be due to some response strategy or sentence processing difficulty.

The critical condition included 50 sentences that tested the participants’ knowledge of different nongeneric uses of the article the. These sentences came in grammatical and ungrammatical versions. In the grammatical condition, test-takers had to repeat the sentence correctly preserving the articles in their target positions. The ungrammatical condition tested whether participants would supply the articles that had been deliberately removed from the sentences in the target positions. Obviously, the grammatical condition had to be easier than the ungrammatical condition because it only required participants to repeat the sentence correctly and to attend to the article provided in the sentence. In contrast, the ungrammatical condition required test-takers to notice the omitted article and to correct the error by supplying the necessary form of the article. The participants’ EI score was computed in the following way: a correctly repeated article in the sentence target position in the grammatical condition or a correctly supplied article in the target position in the ungrammatical condition was given a score of 1; an omission of an article in the sentence target position in the grammatical condition or a failure to supply an article in the target position in the ungrammatical condition received a score of 0. Participants’ cumulative EI scores were subjected to a repeated measures ANOVA with the native language as a between-participants factor and the type of article (5 levels) and grammaticality (2 levels) as within-participants factors. Using an alpha level of 0.05, the analysis yielded a significant interaction between language and grammaticality (F(2, 45) = 5.13, p < 0.01), language and article type (F(8, 180) = 3.8, p < 0.001), and article type and grammaticality (F(4, 180) = 3.74, p < 0.01). There was a significant within-participants main effect of grammaticality (F(1, 45) = 97.42, p < 0.001) and article type (F(4, 180) = 25.64, p < 0.001). The between-participants effect of the native language was also significant (F(2, 45) = 31.53, p < 0.01). Tukey HSD post-hoc tests revealed that while Spanish speakers were not statistically different from the native English speakers in their overall use of the article (p = 0.1), Russian speakers omitted the article in obligatory context significantly more often than either Spanish or English speakers (p < 0.001).

After the omnibus test, we compared the participants’ use of the article in each of the five nongeneric article conditions for each language group. Regarding the grammatical condition, the analysis revealed that the Spanish speakers did not differ significantly from the English native speakers in their use of articles across all five article types. In contrast, when Russian speakers repeated grammatical sentences, they tended to omit the articles significantly more often than either Spanish or English speakers in each of the five article categories despite the fact that the articles were present in the input. With regard to the ungrammatical condition, Spanish speakers were not statistically different from the native speakers in the situational, textual and structural conditions, but they supplied significantly fewer articles in the conventional and cultural conditions. The mean EI score of the Russian speakers in the ungrammatical condition was again significantly lower than that of native speakers of English and Spanish speakers along all article categories. The results of the inter-group comparisons for each grammaticality condition are reported in Table 7.

Table 7.

Inter-group comparisons of the elicited imitation (EI) scores in the critical condition.

Grammaticality	Language	Conventional	Situational	Cultural	Textual	Structural
Grammatical	English	4.94^a	4.81^a	4.88^a	4.81^a	5.00^a
	Spanish	4.31^a	4.63^a	4.25^a	4.81^a	4.88^a
	Russian	3.19^b	3.25^b	3.25^b	3.94^b	4.25^b
Ungrammatical	English	3.56^a	3.94^a	4.06^a	4.38^a	4.63^a
	Spanish	2.44^b	4.13^a	2.94^b	3.56^a	4.19^a
	Russian	1.38^c	1.63^b	1.50^c	1.88^b	2.88^b

Notes. The maximum possible EI score in each article condition is 5.0. Means with a common superscript are not significantly different from each other by the Tukey HSD test at alpha level of 0.05 (e.g. superscript ‘a’ in the grammatical condition for the conventional article use means that the Spanish and the English participants’ mean EI score in the conventional condition was not statistically different from each other, while superscript ‘b’ indicates that the Russian participants’ EI score for the conventional article use was significantly different from that of the Spanish and the English participants; superscript ‘c’ indicates that the mean was statistically different from the means with superscripts ‘a’ and ‘b’).

Within-group comparisons showed that different uses of the definite article the presented varying difficulty across the three language groups, especially in the ungrammatical condition. The analysis of the ungrammatical condition revealed a significant effect of the article type on the EI score in the three language groups (English: F(4, 75) = 3.75, p < 0.05; Spanish: F(4, 75) = 7.11, p < 0.001; and Russian: F(4, 75) = 2.53, p < 0.05). The results are graphically displayed in Figure 4.

Figure 4.

Participants’ EI score across all article categories in the grammatical and ungrammatical conditions.

For the native speaker group, the EI score was the highest in the structural condition and the lowest in the conventional condition. The difference in the mean score between these two conditions was statistically significant (p < 0.01). Situational, cultural, and textual uses occupied the middle position. Although Russian participants scored significantly lower than native English speakers across the five article categories, the overall pattern of the EI score for Russian participants resembled that of the native English speakers. The structural use of the definite article was significantly easier than conventional, situational, and cultural uses (p < 0.05). For the Spanish group, situational and structural uses of the turned out to be the easiest; the EI score in each condition was significantly different from the score in either the conventional or the cultural conditions (p < 0.05).

Finally, in order to check the extent to which the participants were aware of the errors in the EI task and whether they consciously made corrections, the participants’ responses to the immediate recall protocol were analysed. The results indicated that about 94% of Spanish, 88% English, and 69% Russian speakers reported noticing the errors in the sentences (Table 8). However, when the participants were asked what kind of errors they noticed, most participants reported violations of plural number marking. Interestingly, Spanish participants (81.25%) noticed violations in the article use more often than English (56%) and Russian participants (25%). The group difference was statistically significant (F(2, 45) = 6.1, p < 0.05). When asked how often they thought errors occurred in the test, Spanish and English speakers were close in their estimation of the actual overall frequency of errors in the test (46.88% and 45.31%, respectively), whereas Russian speakers underestimated the frequency of error occurrence (only 26.7%). Finally, when asked if they remember correcting errors in their productions during the test, about 94% of English participants, 87.5% of Spanish speakers and only 50% Russian participants remembered correcting errors during the task.

Table 8.

Recall protocol results (percentages).

Language	Reported noticing of errors	Noticing of article violations	Reported error occurrence	Reported error correction
English	87.75	56.00	45.31	93.75
Spanish	93.75	81.25	46.88	87.50
Russian	68.75	25.00	26.70	50.00

V Discussion

The primary goal of the present study was to examine the role of the learners’ native language in their spontaneous, automatic use of the English definite article in its five nongeneric interpretations in an oral elicited imitation (EI) task. Specifically, two groups of L2 learners were examined: highly proficient adult native speakers of Russian (a language which does not have articles) and Spanish (a language with articles). Overall inter-group comparisons of the participants’ elicited imitation performance demonstrated that in the grammatical condition, L1 Spanish participants were as accurate as the native speakers of English in reproducing the definite article in the sentences’ target positions in their elicited repetitions. In contrast, L1 Russian participants showed a tendency to omit the definite article in their sentence productions significantly more often than either English or Spanish speakers even when the task simply required repeating a grammatically correct sentence. With regard to the sentences containing article omissions (ungrammatical condition), L1 Spanish speakers differed from the native control group only in their suppliance of the conventional and cultural the, but were not statistically different in the situational, textual, and structural uses of the article. The degree of suppliance of the nongeneric definite article by L1 Russian participants was significantly lower than that of both Spanish and English speakers along all nongeneric article categories.

We believe that such group differences cannot be due to the groups’ different response strategies or processing difficulties because the analysis of the participants’ performance in the control and filler conditions clearly indicates that both Russian and Spanish speakers consistently make corrections in those conditions, and that their elicited imitation performance in the control and filler conditions is not statistically different. The differences in the EI score between the Spanish and the Russian groups emerge only in the critical (nongeneric article) condition. Moreover, the reported group differences in the use of the nongeneric the are unlikely to be due to learner differences (such as the duration of formal instruction in English, length of residency, proficiency, language exposure, etc.), because the participants were closely matched along these parameters with no significant differences between groups. In fact, except for the length of residency in the USA, the Russian participants had slightly greater values along these parameters. If those factors had affected the participants’ performance, the differences should have led to the opposite pattern from what we observed in the present study: the Russian speakers should have outperformed the Spanish speakers, which is not the case. We therefore argue that such inter-group differences in L2 article use are due to the effect of the native language. Spanish speakers use the definite article in English more consistently because they transfer some of the semantics of the L1 article over to English. Russian lacks articles, so there is no basis for transfer to happen (which is why we refer to Russian and English as being in a zero relation to each other with regard to articles).

Importantly, unlike many previous studies on L2 article use and acquisition, the present study uses a modified oral elicited imitation procedure designed to tap the learners’ implicit knowledge of the English nongeneric article. In light of the methodological approach, the results suggest that L1 transfer effects do not only operate at the level of explicit knowledge, but also apply to learners’ implicit knowledge. The fact that L1 Spanish participants produce definite articles in all nongeneric contexts with higher accuracy than L1 Russian participants suggests that they transfer their implicit knowledge from Spanish to English, and are able to access and retrieve it automatically, routinely and spontaneously. In contrast, L1 Russian learners are not expected to have implicit knowledge of articles that they could transfer from Russian to English because Russian does not grammaticalize definiteness, and there are no similar structures in Russian. As a result, L1 Russian learners of English are faced with the challenge of acquiring an entirely new way of encoding the concept of definiteness that is relevant for the L2. This makes them start off at a lower point of article acquisition and experience more difficulty in using English definite articles in a native-like manner, even at advanced stages of proficiency. This is not to say that they cannot acquire any implicit knowledge of articles. While we believe that implicit knowledge can be learned, there are age-related deficits in the ability to learn an L2 structure implicitly (Ellis, 2009) and to grammaticalize/proceduralize it (Jiang et al., 2011). The lack of implicit, proceduralized knowledge of semantic features associated with the notion of definiteness makes Russian speakers less able to automatically detect violations in definite article use and to spontaneously supply the necessary articles in their oral productions.

The same explanation can be applied to account for the results in the filler condition. Because both Spanish and Russian grammaticalize the concept of plurality, L1 speakers of these languages are sensitive to violations of plurality marking in English, and make spontaneous corrections of plural marking violations in their oral productions in the EI task. Data from the immediate recall protocol provide further evidence in favor of the transfer of implicit knowledge from the speaker’s L1. While both Russian and Spanish participants reported noticing errors in plural noun marking, Spanish speakers noticed violations in the article use more often than Russian participants, and they were also more accurate in their estimation of the overall error occurrence in the test compared to the Russian participants, who underestimated the frequency of article violations. This suggests that the use of the L2 article constitutes part of the automatic, implicit language competence for L1 Spanish speakers, but not for L1 Russian speakers.

While the Spanish speakers clearly have an overall advantage over the Russian speakers in the use of the L2 article, suggesting a positive L1 transfer, their performance on different nongeneric article types was not uniform. Despite different stimulus materials and methodology used in the present study, the Spanish speakers’ use of the nongeneric article closely corresponded to the hierarchy of article difficulty reported in earlier studies by Ansarin (2004), García Mayo (2008), Liu and Gleason (2002), and Wong and Quek (2007). Whereas the situational and the structural conditions were the easiest, the cultural and the conventional uses were the hardest, with the textual use falling in between. Although the Spanish participants performed significantly better than the Russian participants on the cultural and conventional uses, they did not perform as accurately as the native speakers. This observation deserves special attention. Taking into account that Spanish, like English, requires a definite article in all five nongeneric contexts examined in this study, why should the conventional and the cultural article uses be more difficult than other uses? In order to address this issue, we checked whether particular test items in the two conditions elicited more error omissions than other items, but no such items were identified. We also asked five native speakers of Spanish to translate the stimulus items from the conventional and the cultural conditions, having deleted all articles from the sentences beforehand (both definite and indefinite). All native Spanish-speaking informants supplied a definite article in the test items’ target positions. So a mere translation from English into Spanish would have given our Spanish participants the correct form, but, clearly, this is not what they did in the test.

We entertain several explanations of why the cultural and the conventional uses pose more difficulty than other nongeneric article uses. It is possible that, unlike textual, structural and situational uses where the presence of the is always required, there is a greater inter-language variability with regard to the use of the cultural and conventional definite determiner in Spanish and in English. In certain cases, the article in English (e.g. The Hague) directly translates to Spanish (La Haya), but in other cases there is no direct correspondence (e.g. Cairo in English, but El Cairo in Spanish). That is why Spanish speakers may not necessarily assume that if a referent in the cultural or conventional context requires a definite article in Spanish, it will do so in English as well. As suggested by an anonymous reviewer, a higher rate of article omission in the cultural and conventional contexts may also reflect the article’s ‘redundancy’ in these contexts, especially with proper nouns, because they are always definite, even when no definite article is present. Notably, even the native speakers in the study do not supply articles in the cultural and conventional contexts with a 100% accuracy (Figure 4, ungrammatical condition), suggesting that they tolerate some degree of article optionality in the cultural and conventional contexts. This relates to another possible explanation of why Spanish speakers of English supply fewer articles in the conventional and cultural contexts: cultural and conventional uses of the are unpredictable and cannot be deduced from general rules of article use. For example, while the is used with some cultural nouns, it is absent in other NPs, even though the noun may refer to the same referent (e.g. the Netherlands but Holland). Similarly, while the is used with musical instruments (e.g. play the piano), it is not used with sports (e.g. play soccer). Therefore, in order to learn when to use the in cultural and conventional contexts, Spanish learners must learn how these uses differ in the two languages on a case-by-case basis and memorize these instances through L2 cultural experience and exposure.

With regard to the L1 Russian group, conventional and cultural uses of the nongeneric the also presented difficulty, though no more so than situational uses. Textual and structural uses of the seemed to present less difficulty for Russian speakers and yielded a higher EI score. In order to account for the observed pattern of results, we need to remember that situational the is used when the referent is a first-mention noun, which can be sensed directly or indirectly as definite or unique by interlocutors. In other words, in order to use situational the correctly, one needs to somehow encode the meaning of definiteness and rely on one’s perception of definiteness of a given referent because there are no other cues in the immediate linguistic environment indicating that a definite article needs to be used. In contrast, when the is used in its textual or structural interpretations, the speaker can rely on additional cues derived from either immediate or more distant context. In the latter case, the use of the can be derived from an easily deducible explicit rule drawing on the structural constraints of the article use; for example: ‘if a noun is referred to for the second time, use the’ (textual), or ‘if a noun has a modifier, use the’ (structural). The finding that L1 Russian participants perform better in textual and structural article conditions as opposed to situational, cultural, and conventional conditions suggests that the Russian speakers may have formulated an explicit strategy that they have been able to automatize/proceduralize through practice, and can compensate – at least to a certain extent – for their lack of grammaticalized semantic knowledge of different types of articles by taking into account syntactic and structural cues to article use.

Overall, our results suggest that advanced Russian learners of English have knowledge of the syntactic distribution of articles, but still experience difficulties with the semantic aspects of articles. That is why we see a considerable difference in Spanish and Russian participants’ use of situational the: L1 Spanish speakers of English transfer a great part of article semantics from L1 to L2 and so they correctly categorize English articles on the basis of definiteness, whereas Russian speakers fail to do so because Russian speakers lack relevant, transferrable semantics. Instead, Russian speakers resort to structural and/or discourse means to determine definiteness of a referent.

In conclusion, our findings demonstrate that L1 Spanish learners of English – speakers of the language with a complex article system – use the English nongeneric definite article with almost native-like accuracy, while Russian learners – whose native language does not have articles – have a greater tendency to omit articles in their oral production. A comparison of the performance of the Spanish and Russian speakers on different types of the English nongeneric definite article, however, shows that different article contexts do not present equal difficulty for L2 learners, and that learners from different L1s employ different strategies for determining the use of the L2 article. This study offers useful insights and helps to uncover the underlying causes of article difficulty for learners from different L1s.

Footnotes

Appendix 1. Acknowledgements

The authors would like to thank the anonymous Second Language Research reviewers for their valuable feedback and many insightful comments.

Declaration of conflicting interest

The author declares that there is no conflict of interest.

Funding

Funding for this study was provided by the School of Languages, Literatures, and Cultures at the University of Maryland, USA.

Authors’ Note

Preliminary results were presented at the 21st Annual Conference of the European Second Language Association (EUROSLA).

References

Ambridge

Pine

(2006) Testing the agreement/tense omission model using an elicited imitation paradigm. Journal of Child Language 33: 879–98.

Ansarin

(2004) Nongeneric use of the definite article ‘the’ by Persian learners. Journal of Faculty of Letters and Humanities 47: 1–11.

Avery

Radišić

(2007) Accounting for variability in the acquisition of English articles. In: Belikova

Meroni

Umeda

(eds) Proceedings of the 2nd Conference on Generative Approaches to Language Acquisition, North America (GALANA). Somerville, MA: Cascadilla Proceedings Project, 1–11.

Bley-Vroman

Chaudron

(1994) Elicited imitation as a measure of second-language competence. In: Tarone

Gass

Cohen

(eds) Research methodology in second-language acquisition. Mahwah, NJ: Lawrence Erlbaum, 245–61.

Brown

(1980) Relative merits of four methods for scoring cloze tests. Modern Language Journal 64: 311–17.

Butler

(2002) Second language learners’ theories on the use of English articles: An analysis of the metalinguistic knowledge used by Japanese students in acquiring the English article system. Studies in Second Language Acquisition 24: 451–80.

Celce-Murcia

Larsen-Freeman

(1999) The grammar book: An ESL teacher’s course. 2nd edition. Boston, MA: Heinle and Heinle.

Christian

(1961) Some consequences of the lack of a definite and indefinite article in Russian. The Slavic and East European Journal 5: 1–11.

Christophersen

(1939) The articles: A study of their theory and use in English. Copenhagen: Einar Munksgaard.

10.

Corder

(1967) The significance of learners’ errors. International Review of Applied Linguistics 5: 161–70.

11.

Corder

(1981) Idiosyncratic dialects and error analysis. International Review of Applied Linguistics 9: 147–59.

12.

Dulay

Burt

Krashen

(1982) Language two. Oxford: Oxford University Press.

13.

Eadie

Fey

Douglas

Parsons

(2002) Profiles of grammatical morphology and sentence imitation in children with specific language impairment and Down syndrome. Journal of Speech, Language, and Hearing Research 45: 720–32.

14.

Ekiert

(2004) Acquisition of the English article system by speakers of Polish in ESL and EFL settings. Teachers College, Columbia University Working Papers in TESOL and Applied Linguistics 4: 1–23.

15.

Ellis

(1994) The study of second language acquisition. Oxford: Oxford University Press.

16.

Ellis

(2000) Task-based research and language pedagogy. Language Teaching Research 4: 193–220.

17.

Ellis

(2009) Measuring implicit and explicit knowledge of a second language. In: Ellis

Loewen

Elder

Erlam

Philp

Reinders

(eds) Implicit and explicit knowledge in second language learning and teaching. Bristol: Multilingual Matters, 31–64.

18.

Ellis

Loewen

Erlam

(2006) Implicit and explicit corrective feedback and the acquisition of L2 grammar. Studies in Second Language Acquisition 28: 339–69.

19.

Erlam

(2006) Elicited imitation as a measure of L2 implicit knowledge: An empirical validation study. Applied Linguistics 27: 464–91.

20.

Erlam

(2009) The elicited oral imitation test as a measure of implicit knowledge. In: Ellis

Loewen

Elder

Erlam

Philp

Reinders

(eds) Implicit and explicit knowledge in second language learning and teaching. Bristol: Multilingual Matters, 65–93.

21.

García Mayo

del

(2007) Investigating tasks in formal language learning. Clevedon: Multilingual Matters.

22.

García Mayo

del

(2008) The acquisition of four nongeneric uses of the article the by Spanish EFL learners. System 36: 550–65.

23.

García Mayo

del

(2009) Article choice in L2 English by Spanish speakers: Evidence for full transfer. In: García Mayo

del

Hawkins

(eds) Second language acquisition of articles: Empirical findings and theoretical implications. Amsterdam: John Benjamins, 13–36.

24.

Gass

Selinker

(1992) Language transfer in language learning. Revised edition. Amsterdam: John Benjamins.

25.

Goad

White

(2004) Ultimate attainment of L2 inflections: Effects of L1 prosodic structure. In: Foster-Cohen

Ota

Sharwood Smith

Sorace

(eds) EUROSLA Yearbook. Amsterdam: John Benjamins, 119–45.

26.

Goad

White

(2009). Prosodic transfer and the representation of determiners in Turkish-English interlanguage. In: Snape

Leung

Sharwood Smith

(eds) Representational deficits in SLA: Studies in honor of Roger Hawkins. Amsterdam: John Benjamins, 1–26.

27.

Goad

White

Steele

(2003) Missing inflection in L2 acquisition: Defective syntax or L1-constrained prosodic representations? Canadian Journal of Linguistics 48: 243–63.

28.

Graham

Lonsdale

Kennington

Johnson

McGhee

(2008) Elicited imitation as an oral proficiency measure with ASR scoring. Proceedings of the LREC (Language Resources and Evaluation) Conference, 1604–10.

29.

Granena

(2009) Elicited imitation as a measure of implicit L2 knowledge and working memory capacity. Unpublished paper presented at Second Language Research Forum Annual Conference, Michigan State University, MI.

30.

Hamayan

Saegert

Laraudee

(1977) Elicited imitation in second language learners. Language and Speech 20: 86–97.

31.

Hawkins

(1978) Definiteness and indefiniteness. London: Croom Helm.

32.

Hawkins

Al-Eid

Almahboob

. (2006) Accounting for English article interpretation by L2 speakers. In: Foster-Cohen

Medved Krajnovic

Mihaljević Djigunović

(eds) EUROSLA Yearbook 6. Amsterdam: John Benjamins, 7–25.

33.

Hirata-Edds

(2006) Influence of second language immersion in Cherokee on children’s development of past tense in their first language, English. Unpublished PhD thesis, University of Lawrence, KS.

34.

Ionin

(2003) Article semantics in second language acquisition. Unpublished PhD thesis, Massachusetts Institute of Technology, Cambridge, MA.

35.

Ionin

Montrul

(2010) The role of L1 transfer in the interpretation of articles with definite plurals in L2 English. Language Learning 60: 877–925.

36.

Ionin

Wexler

(2004) Article semantics in L2-acquisition: the role of specificity. Language Acquisition 12: 3–69.

37.

Ionin

Zubizarreta

Maldonado

(2008) Sources of linguistic knowledge in the second language acquisition of English articles. Lingua 118: 554–76.

38.

Jarvis

(2000) Methodological rigor in the study of transfer: Identifying L1 influence in the interlanguage lexicon. Language Learning 50: 245–309.

39.

Jarvis

(2002) Topic continuity in L2 English article use. Studies in Second Language Acquisition 24: 387–418.

40.

Jensen

Vinther

(2003) Exact repetition as input enhancement in second language acquisition. Language Learning 5: 373–428.

41.

Jespersen

(1949) A modern English grammar on historical principles, 7. Copenhagen: Munksgaard.

42.

Jiang

Novokshanova

Masuda

Wang

(2011) Morphological congruency and the acquisition of L2 morphemes. Language Learning 61: 940–67.

43.

Kellerman

(1995) Crosslinguistic influence: Transfer to nowhere? Annual Review of Applied Linguistics 15: 125–50.

44.

Kellerman

Sharwood-Smith

(1986) Cross-linguistic influence in second language acquisition. London: Pergamon.

45.

Leonetti

(1999) El artículo [The Article]. In: Bosque

Demonte

(eds) Gramática descriptiva de la lengua española [Descriptive grammar of Spanish]. Madrid: Espasa, 787–890.

46.

Leung

(2002) Functional categories in second and third language acquisition: A crosslinguistic study of the acquisition of English and French by Chinese and Vietnamese speakers. Unpublished PhD thesis, McGill University, Montreal, QC, Canada.

47.

Liu

Gleason

(2002) Acquisition of the article the by nonnative speakers of English: An analysis of four nongeneric uses. Studies in Second Language Acquisition 24: 1–26.

48.

Markman

Spilka

Tucker

(1975) The use of elicited imitation in search of an interim French grammar. Language Learning 75: 31–41.

49.

Master

(1997) The English article system: Acquisition, function, and pedagogy. System 25: 215–32.

50.

Master

(2002) Information structure and English article pedagogy. System 30: 331–48.

51.

Mizuno

(1999) Interlanguage analysis of the English article system: Some cognitive constraints facing the Japanese adult learners. International Review of Applied Linguistics 37: 127–52.

52.

Montrul

Ionin

(2010) Transfer effects in the interpretation of definite articles by Spanish heritage speakers. Bilingualism: Language and Cognition 13: 449–73.

53.

Munnich

Flynn

Martohardjono

(1994) Elicited imitation and grammaticality judgment tasks: What they measure and how they relate to each other. In: Tarone

Gass

Cohen

(eds) Research methodology in second-language acquisition. Mahwah, NJ: Lawrence Erlbaum, 227–45.

54.

Odlin

(1989) Language transfer: Cross-linguistic influence in language learning. Cambridge: Cambridge University Press.

55.

Papadopoulou

Clahsen

(2003) Parsing strategies in L1 and L2 sentence processing: A study of relative clause attachment in Greek. Studies in Second Language Acquisition 24: 501–28.

56.

Parrish

(1987) A new look at methodologies in the study of article acquisition for learners of ESL. Language Learning 37: 361–83.

57.

Ringbom

(1987) The role of the mother tongue in foreign language learning. Clevedon: Multilingual Matters.

58.

Ringbom

(1990) Effects of transfer in foreign language learning. In: Dechert

(ed.) Current trends in European second language acquisition research multilingual matters. Clevedon, 205–18.

59.

Ringbom

(2007) Cross-linguistic similarity in foreign language learning. Clevedon: Multilingual Matters.

60.

Robertson

(2000) Variability in the use of the English article system by Chinese learners of English. Second Language Research 16: 135–72.

61.

Schachter

(1983) A new account of language transfer. In: Gass

Selinker

(eds) Language transfer in language learning. Rowley, MA: Newbury House, 98–111.

62.

Schwartz

Sprouse

(1996) L2 cognitive states and the Full Transfer/Full Access model. Second Language Research 12: 40–72.

63.

Sidorova

(2006) Lexical means of expressing the category of definiteness / indefiniteness in the Russian language (in comparison with the Italian language). Unpublished PhD thesis, Novosibirsk State University, Russia.

64.

Snape

García Mayo

del

Gürel

(2009) Spanish, Turkish, Japanese and Chinese L2 learners’ acquisition of generic reference. In: Bowles

Ionin

Montrul

Tremblay

(eds) Proceedings of the 10th Generative Approaches to Second Language Acquisition Conference (GASLA 2009). Somerville, MA: Cascadilla Proceedings Project, 1–8.

65.

Tarone

Parrish

(1989) Task-related variation in interlanguage: The case of articles. Language Learning 38: 21–44.

66.

Thomas

(1989) The acquisition of English articles by first- and second-language learners. Applied Psycholinguistics 10: 335–55.

67.

Trenkic

(2002) Form–meaning connections in the acquisition of English articles. EUROSLA Yearbook 2: 115–33.

68.

Trenkic

(2004) Definiteness in Serbian/Croatian/Bosnian and some implications for the general structure of the nominal phrase. Lingua 114: 1401–27.

69.

Trenkic

(2007) Variability in second language article production: beyond the representational deficit vs. processing constraints debate. Second Language Research 23: 289–327.

70.

Tryzna

(2009) Questioning the validity of the Article Choice Parameter and the Fluctuation Hypothesis: Evidence from L2 English article use by L1 Polish and L1 Mandarin Chinese speakers. In: García Mayo

del

Hawkins

(eds) Second language acquisition of articles. Empirical findings and theoretical implications. Amsterdam: John Benjamins, 67–86.

71.

Vargas-Barón

(1952) The function of the definite article in Spanish. Hispania 35: 410–14.

72.

Vinther

(2002) Elicited imitation: A brief overview. International Journal of Applied Linguistics 12: 54–73.

73.

White

(2009) Accounting for L2-English learners’ article choices. MSU Working Papers in SLS 1: 14–37.

74.

Wong

Quek

(2007) Acquisition of the English definite article by Chinese and Malay ESL learners. Electronic Journal of Foreign Language Teaching 4: 210–34.

75.

Young

(1995) Form-function relations in articles in English interlanguage. In: Bayley

Preston

(eds) Second language acquisition and linguistic variation. Amsterdam: Benjamins, 135–75.